融合非局部空间灰度信息的三维Otsu法

为了提高最大类间方差阈值分割法(Otsu)对于图像噪声的鲁棒性,提出融合非局部空间灰度信息的三维Otsu法。该方法利用图像像素的灰度信息、邻域中值灰度信息和非局部空间灰度信息进行直方图统计,构建新颖的三维直方图,采用最大类间方差作为阈值选取准则。实验结果表明新方法对于噪声的鲁棒性要优于原始三维Otsu法,能够获得更加令人满意的分割结果。

Three-dimensional Otsu's method with non local spatial gray information

In order to improve the robustness of the maximum between-cluster variance thresholding segmentation method(Otsu) to image noise, a three-dimensional Otsu’s method with the non local spatial gray information is proposed. The gray information, the neighboring median gray information and the non local spatial gray information of the pixel are utilized to perform histogram statistics. A novel three-dimensional histogram is constructed. The maximum between-cluster variance is adopted as the threshold selection rule. Experimental results show that the proposed method is more robust than the traditional three-dimensional Otsu’s method and can obtain more satisfying segmentation performance.